Possible Regulation of the Conventional Calpain System by Skeletal Muscle-specific Calpain, p94/Calpain 3*

p94 (also called calpain 3) is the skeletal muscle-specific calpain and is considered to be a “modulator protease” in various cellular processes. Analysis of p94 at the protein level is an urgent issue because the loss of p94 protease activity causes limb-girdle muscular dystrophy type 2A. In this study, we enzymatically characterized one alternatively spliced variant of p94, p94:exons 6–15–16– (p94Δ), which lacks two of the p94-specific insertion sequences. In contrast to p94, which has hardly been studied enzymatically due to its rapid, thorough, and apparently Ca2+-independent autolytic activity, p94Δ was stably expressed in COS and insect cells. p94Δ showed Ca2+-dependent caseinolytic and autolytic activities and an inhibitor spectrum similar to those of the conventional calpains. However, calpastatin did not inhibit p94Δ and is a substrate for p94Δ, which is consistent with the properties of p94, presenting p94 as a possible regulator of the conventional calpain system. We also established a semi-quantitative fluorescence resonance energy transfer assay using the calpastatin sequence specifically to measure p94 activity. This method detects the activity of COS-expressed p94 and p94Δ, suggesting that it has potential to evaluate p94 activity in vivo and in the diagnosis of limb-girdle muscular dystrophy type 2A.

p94 (also called calpain 3) is the skeletal muscle-specific calpain and is considered to be a "modulator protease" in various cellular processes. Analysis of p94 at the protein level is an urgent issue because the loss of p94 protease activity causes limb-girdle muscular dystrophy type 2A. In this study, we enzymatically characterized one alternatively spliced variant of p94, p94:exons 6 ؊ 15 ؊ 16 ؊ (p94⌬), which lacks two of the p94-specific insertion sequences. In contrast to p94, which has hardly been studied enzymatically due to its rapid, thorough, and apparently Ca 2؉ -independent autolytic activity, p94⌬ was stably expressed in COS and insect cells. p94⌬ showed Ca 2؉ -dependent caseinolytic and autolytic activities and an inhibitor spectrum similar to those of the conventional calpains. However, calpastatin did not inhibit p94⌬ and is a substrate for p94⌬, which is consistent with the properties of p94, presenting p94 as a possible regulator of the conventional calpain system. We also established a semi-quantitative fluorescence resonance energy transfer assay using the calpastatin sequence specifically to measure p94 activity. This method detects the activity of COS-expressed p94 and p94⌬, suggesting that it has potential to evaluate p94 activity in vivo and in the diagnosis of limb-girdle muscular dystrophy type 2A.
Calpain (EC 3.4.22.17, clan CA, family C2) is a Ca 2ϩ -requiring cysteine protease representing one of the most important families of the cysteine proteases (1)(2)(3)(4)(5)(6)(7)(8)(9). To date, various molecules showing significant similarity to the calpain protease domain have been identified in almost all kinds of living organisms and constitute the "calpain superfamily" (6). Two representative members,and m-calpains, the so-called "conventional" calpains, are ubiquitously expressed and have been well characterized. These two calpains consist of a distinct larger catalytic subunit containing a protease domain (-or m-calpain large subunit, abbreviated as CL 1 or mCL, respectively) and a common smaller regulatory subunit (abbreviated as 30K according to its molecular weight). On the basis of amino acid similarities, the large and small subunits have been described as consisting of four and two domains, respectively, which agrees with the recently resolved three-dimensional structure of m-calpain (10,11) (Fig. 1A).
Conventional calpain has a specific endogenous proteinaceous inhibitor, calpastatin (12). Calpastatin contains four repetitive inhibitory units, each of which inhibits equimolar amounts of conventional calpain. The conserved reactive site interacts with the calpain protease domain, whereas the flanking ␣-helical regions bind to domains IV and VI of the large and small subunits, respectively. Synthetic oligopeptides (see Fig.  4D) corresponding to the calpastatin-reactive site specifically inhibit conventional calpain efficiently, although their inhibitory activity is weaker than that of the full-length inhibitory unit.
The primary structure of p94 (also called calpain 3) is very similar to those of CL and mCL throughout the entire molecule, beyond the p94-specific sequences NS, IS1, and IS2 (13). Previous studies have revealed several unique characteristics of p94 that diverge greatly from those of the conventional calpains. For instance, 1) p94 undergoes very rapid, thorough, and apparently Ca 2ϩ -independent autolysis in solution (halflife in vitro is less than 10 min) (14); 2) inhibitors of the conventional calpains, including calpastatin, EDTA, and EGTA, have no effect on p94 autolysis (14); 3) the gene for p94 produces several alternatively spliced products (15,16); and 4) p94 associates with the N2 line region and the C terminus of connectin/titin, the gigantic filamentous molecule essential for myofibrils (17,18; for a review of connectin/titin, see Refs. * This work was supported in part by a grant-in-aid for scientific research on priority areas (cell cycle) from the Ministry of Education, Science, Sports and Culture, a grant-in-aid for scientific research and research fellowships for young scientists from the Japan Society for the Promotion of Science, Research Grant 14B-4 for Nervous and Mental Disorders from the Ministry of Health, Labor and Welfare, the Deutsch Forschungsgemeinschaft Grant La668/7-1, and "Ground-based Research Announcement for Space Utilization" promoted by the Japan Space Forum. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The  19 -21). The predominant expression of p94 in skeletal muscle, where its mRNA levels are ϳ10 times higher than those of CL and mCL, indicates the physiological importance of p94 in that tissue (13). Consistent with this, a defect in the p94 gene causes limb-girdle muscular dystrophy type 2A (LGMD2A), suggesting that p94 functions are indispensable for proper muscle functions (22). Several studies, including ours, indicate that the loss of substrate processing activity, but not hyperactivation or a defect in the structural properties of p94, causes LGMD2A (16,(23)(24)(25)(26). Therefore, it has become an urgent issue to determine the in vivo substrates of p94 to gain insight into the physiological functions of p94 and its relationships to molecular mechanisms of LGMD2A (27).
However, the autolytic activity of p94 has hampered the study of the p94 protein. Therefore, we have focused on identifying the conditions that will allow us to analyze the proteolytic activity of p94. We have observed previously that deletion of either the IS1 or IS2 region, which is mainly encoded by exon 6 or exons 15 ϩ 16, respectively, prevents the rapid autolysis of p94 (14). In accord with this, we have recently shown that alternatively spliced variants of p94 that lack either exon 6, exons 15 ϩ 16, or all of these are detected as non-autolyzed forms, unlike the full-length p94, when they are recombinantly expressed in COS cells (16).
In this study, we chose one of the splicing variants, p94:exons 6 Ϫ 15 Ϫ 16 Ϫ (abbreviated as p94⌬ in this paper, see Fig. 1A), which is produced from a transcript lacking exons 6, 15, and 16 and therefore lacks most of the p94-specific insertion sequences, IS1 and IS2, for in vitro enzymatic characterization on the basis of its stability and expression efficiency. The enzymatic properties of p94⌬ were compared with those of p94, and and m-calpains. We thus identified several novel enzymatic properties of p94. Notably, neither p94 nor p94⌬ is inhibited by calpastatin, and moreover, both hydrolyze calpastatin. These results led us to predict that p94 participates in the regulation of the conventional calpain-calpastatin system in skeletal muscle. We used these findings to establish a calpastatin-based fluorescence assay specifically to measure p94 protease activity, showing a possibility of application of this assay as a diagnostic tool for LGMD2A.
p94-specific anti-pNS (antigen, PTVISPTVAPRTGAEPRS, at the N terminus) and anti-pIS2 (identical to anti-pK-rich antigen, NTISVDRPVKKKKNKPIIFV, overlapping with the IS2 region) antisera have been described previously (14). Monoclonal antibodies 2C4 and 12A2, which recognize N terminus and the region around the catalytic residue Asn-358 of p94, respectively, were purchased from NovoCastra (Newcastle, UK). Monoclonal antibodies that recognize CL and mCL were purchased from Biomol and Chemicon, respectively. Recombinant domain 1 of human calpastatin protein and the peptides corresponding to its reactive site were purchased from Takara Shuzo (7316 and SP007) and Sigma (C9181).
Recombinant Proteins Expressed in Insect Cells-The cDNA for p94⌬ or p94:C129S, an active-site mutant, was inserted into the pFastBac1 vector using appropriate restriction enzyme sites. Recombinant baculovirus was generated according to the manufacturer's instructions.
Sf-9 cells were cultured and infected as described previously (28). The cells were then shaken vigorously at 27°C for 50 h at a density of 1 ϫ 10 6 cells/ml and collected by centrifugation at 600 ϫ g for 5 min at 4°C. Recombinant proteins were purified following the procedures described previously (28), with minor modifications as follows. The harvested cells were washed twice with ice-cold phosphate-buffered saline. The pellets were suspended in buffer A (20 mM Tris-Cl (pH 7.5), 5 mM EDTA, 10 mM 2-mercaptoethanol (2-ME)) containing 0.1 mM leupeptin, and the cells were homogenized with a French press (SLM Aminco). The cell lysate was then ultracentrifuged at 150,000 ϫ g for 40 min at 2°C. Recombinant proteins were purified from the supernatant by the AKTA system (Amersham Biosciences) and placed in a cold chamber, using successive rounds of column chromatography: DEAE-Toyopearl (Tosoh, Tokyo, Japan), MonoQ (Amersham Biosciences), HiLoad 16/60 Superdex-200, and MonoQ. Fractions containing recombinant proteins were determined by Western blotting analysis and caseinolytic assays (for p94⌬). Recombinantand m-calpains were prepared in the manner described in Ref 28. Bacterial Expression and Purification of Recombinant Proteins-cDNA fragments corresponding to two different N2A regions of connectin/titin, IgI80-81, and IgI82-83, were amplified by PCR from human skeletal muscle cDNA and cloned into the pET vector. Proteins were expressed in Escherichia coli BL21(DE3) and purified as described previously (29).
A bacterial expression vector for the fusion protein generated from enhanced cyan and yellow fluorescence proteins (ECFP and EYFP, respectively) linked by a part of rabbit calpastatin domain 1 (residues 175-255, GenBank TM accession number A26615), Y-C-CSTN, was created by fusing the PCR-amplified coding sequences of ECFP, EYFP, and calpastatin and cloning the construct into the pET16b vector. The recombinant protein, Y-C-CSTN, expressed in E. coli. BL21(DE3), was purified by His-Bind Quick 900 Cartridges (Novagen), according to the manufacturer's protocol. Protein fractions eluted from the cartridge were desalted on a PD-10 column (Amersham Biosciences) and concentrated by Centricon (Millipore, MA).
Protease Activity Assay-Caseinolytic activity of p94⌬ was measured essentially according to Ref. 30. In brief, the composition of the standard assay solution was 100 mM Tris-Cl (pH 7.5), 25 mM 2-ME, 10 mM CaCl 2 . CaCl 2 was replaced with 10 mM EDTA for the control reaction. The reaction was carried out by incubating either p94⌬ (3 g), -calpain (0.2 g), or m-calpain (0.4 g) in 50 l of the standard assay solution containing 3 mg/ml casein under the conditions indicated. The reaction was stopped by the addition of 150 l of 7% (w/v) trichloroacetic acid. The tubes were incubated on ice for 20 min and centrifuged at 15,000 ϫ g for 10 min, and the A 280 of the supernatant was measured. An increase in A 280 of 1.0 in 1 h was defined as 1 unit of caseinolytic activity.
The effect of Ca 2ϩ on p94⌬ was determined under the assay conditions described above, varying either the concentration of Ca 2ϩ , temperature, or pH. Tris acetate buffer was used instead of Tris-Cl buffer to measure the effect of pH. Protease inhibitors were added to the standard assay solutions at the concentrations indicated. The reactions were carried out at 37°C for 45 min, and the caseinolytic activity was determined as described. The effect of each inhibitor was evaluated relative to the inhibitory effect of 10 mM EDTA set to 100.
To determine the substrate hydrolyzing activity of p94⌬, various proteins were incubated with p94⌬ in the standard assay solution. Hydrolysis of the proteins was examined by SDS-PAGE followed by Coomassie Brilliant Blue (CBB) staining or Western blotting.
In the proteolytic assay using Y-C-CSTN as substrate, 2 g of Y-C-CSTN was incubated with p94⌬ in the presence of 15 mM Ca 2ϩ for 0 -45 min at 37°C in a solution containing 0.1 M Tris-Cl (pH 7.5) and 10 mM 2-ME in a volume of less than 20 l. The reaction was stopped by the addition of 0.5 ml of 0.1 M EDTA (pH 8.0) or an equal volume of 2ϫ SDS sample buffer. The fluorescence of the samples was measured by excitation at 434 nm (10 nm bandwidth), and the emission spectra were collected from 450 to 600 nm using an RF-1500 spectrofluorophotometer (Shimadzu, Kyoto, Japan). The ratio of fluorescence at 477 nm to that at 527 nm was used to define proteolytic activity. To profile the dose-dependent changes in fluorescent signals, 0 -2 g of p94⌬ was incubated with Y-C-CSTN (0.6 g) in a buffer consisting of 28 mM Ca 2ϩ , 1 mg/ml bovine serum albumin (BSA), 0.1 mg/ml calpastatin (CSTN) domain 1, 20 mM 2-ME, and 100 mM Tris-Cl (pH 7.5) in a total volume of 15 l for 5-45 min.
To detect p94 activity in the cell lysate, the following protease inhibitors were added to the standard assay mixture described above to inhibit the conventional calpains and other proteases: 10 M Z-D-CH2-DCB, 10 M YVAD-CMK, 1 mM AEBSF, 2 M calpastatin peptide C9181, and 2 mM PMSF. The same number of COS-7 cells (2-3 ϫ 10 5 ) transfected with each plasmid was incubated with Y-C-CSTN in the presence of 5 mM Ca 2ϩ or 10 mM EDTA in a total volume of 20 l for 90 min at 37°C, after which 0.5 ml of 0.1 M EDTA was added. To detect the autolysis of p94 and p94⌬ by Western blotting, the reaction was stopped by the addition of half a volume of 3ϫ SDS sample buffer.
Other Protein Experiments-SDS-PAGE, Western blotting, and subsequent procedures were performed as described previously (14). Proteins were transferred onto polyvinylidene difluoride membranes (Immobilon-P, Millipore Inc.), as recommended by the supplier, and incubated with specific antisera as indicated. The secondary antibody was horseradish peroxidase-coupled goat anti-rabbit or goat anti-mouse IgG (Nichirei Inc., Tokyo, Japan). The antibody complexes were visualized using peroxidase substrate (POD Immunostaining Kit; Wako Inc., Osaka, Japan). To examine the autolytic activity of p94⌬, 1 g of p94⌬ was incubated in 15 l of solution containing 100 mM Tris-Cl (pH 7.5), 25 mM 2-ME, and 10 mM CaCl 2 , at either 0 or 30°C for the indicated times. The reaction was stopped by the addition of an equal volume of 2ϫ SDS sample buffer and subjected to SDS-PAGE followed by CBB staining or Western blotting, as described above.
Protein concentrations were measured using a DC Protein Assay Kit (Bio-Rad) with BSA as the standard. The N-terminal amino acid sequence of the protein fragments was determined on an Applied Biosystems protein sequencer type 477A/120A, according to manufacturer's instructions and the method described previously (31), with some modification.

RESULTS
Expression and Purification of p94⌬ and p94:C129S-Recombinantly expressed p94⌬ and p94:C129S (a protease-inactive point mutant of p94/calpain 3) were purified by four successive rounds of column chromatography, as described under "Experimental Procedures" (see Fig. 1A for structures of p94⌬ and p94:C129S). The purity of these products was estimated to be more than 90% based on a CBB-stained gel after SDS-PAGE, as shown in Fig. 1B. The N-terminal sequence of purified p94⌬ revealed that, after 2 weeks storage at 2°C, most of the purified p94⌬ had been hydrolyzed on the N terminus of Ala-34 (Fig. 1A, arrow [4]) and that anti-pNS antiserum, raised against a peptide corresponding to N-terminal residues 2-19 (14), no longer recognized the purified protein. The p94⌬ sample was recognized by anti-pNS antiserum just after purification. p94:C129S, which was purified by a procedure identical to that used for p94⌬, showed no hydrolysis at the N terminus. Therefore, we concluded that the hydrolysis of p94⌬ was autolytic. This autolysis occurred in the presence of 5 mM EDTA, even at 2°C, which is consistent with our previous observation that p94 rapidly undergoes autolysis in the presence of excess EDTA (14).
The yields of totally purified p94⌬ and p94:C129S proteins were ϳ1.2 and 2.4 mg/liter culture, respectively. Although the N-terminal autolysis of p94⌬ did not require Ca 2ϩ , p94⌬ showed Ca 2ϩ -dependent caseinolytic activity (see Fig. 2; details are discussed later). Analysis of the degraded casein by SDS-FIG. 1. Schematic structure of p94: C129S, p94⌬, and purified proteins. A, schematic structure of p94:C129S and p94⌬. NS, IS1, and IS2 are specific sequences found in p94 but not in the conventional (and m-) calpain large subunits (CL and mCL). Domains I, III, and V are ␣-helical, C2-like Ca 2ϩ -binding, and Gly-rich hydrophobic domains, respectively. Domains IV and VI contain five EF-hand motifs and are very similar to each other. Subdomains IIa and IIb, kept apart in the inactive state, compose the protease domain when activated. Thick horizontal bars indicate the positions of epitopes for the antibodies used in this study. Note that a part of the epitope for anti-pIS2 is retained in p94⌬. Arrows [1] to [4] indicate the proteolytic sites determined in this study. B, purified p94: C129S and p94⌬ used in this study. Lane 1, molecular weight marker; lanes 2 and 3, p94:C129S (5 g) and p94⌬ (5 g PAGE indicated that p94⌬ degrades casein in a manner similar to the caseinolysis by the conventional calpains (data not shown). The specific activity of p94⌬ determined by caseinolysis was 9.0 units/mg, which is ϳ1/80 of the value for recombinant human m-calpain determined with the same expression and purification systems (28,32,33). However, because most of the purified p94⌬ underwent N-terminal autolysis as described above, we concluded that very low specific caseinolytic activity is one of the enzymatic properties of p94⌬ and does not result from partial denaturation or aggregation of the protein.
Enzymatic Properties of p94⌬: Effects of Temperature, pH, Ca 2ϩ Concentration, and Protease Inhibitors-The caseinolytic activity of p94⌬ was examined while varying either temperature, pH, or Ca 2ϩ concentration and compared with the corresponding parameters for recombinantand m-calpains. Fig.  2A shows the temperature dependence of p94⌬ caseinolytic activity. The initial rate of p94⌬ activity increased with temperature. In the range of temperatures examined here, the activity of p94⌬ increased persistently with time up to 60 min. The optimal temperature for 20 min of incubation of p94⌬ was about 42°C, which contrasts with the decrease in activity observed for bothand m-calpains at temperatures above 25°C (Fig. 2B). Although p94⌬ shows maximum activity at around pH 7.5, as do other calpains, p94⌬ maintains its activity even at pH 10 (Fig. 2C). Ca 2ϩ dependence showed that a pCa for half-maximal activity for p94⌬ was 2.9, which is lower than those ofand m-calpains (3.7 and 3.3, respectively) (Fig. 2D).
The effects of various inhibitors on the activity of p94⌬ and and m-calpains were investigated (Table I). p94⌬ has a spectrum of inhibitors very similar to those ofand m-calpains, except calpastatin. The calpastatin fragment inhibited and m-calpains completely at molar ratios of almost 1:1, whereas p94⌬ was not inhibited by a 100-fold molar excess of the calpastatin fragment or calpastatin-related peptides (Fig.  2E). In fact, calpastatin was shown to be a substrate for p94⌬ (see below). Because the IS1 region is a target for p94 autolysis (31,34), we examined the possibility that synthetic peptides corresponding to IS1 would have a competitive inhibitory effect on p94. However, none of the peptides inhibited the caseinolytic activity of p94⌬, even at a molar ratio of 100:1 (data not shown).
Autolytic Activity of p94⌬-p94⌬ is much more stable than full-length wild-type p94. However, as described above, limited autolysis at the N terminus of p94⌬ occurs even in the presence of EDTA at 2°C. Therefore, the autolytic activity of p94⌬ was analyzed in more detail. p94⌬, already lacking the N-terminal 33 amino acids of p94, underwent further autolysis in the presence of Ca 2ϩ . As shown in Fig. 2, F and G, this second phase autolysis generated an 83-kDa fragment very rapidly (t1 ⁄2 ϭ 5 min at 30°C and 20 min at 0°C). No change in caseinolytic activity or Ca 2ϩ dependence was caused by preincubation of p94⌬ for 5, 10, 20, or 60 min at 30°C (data not shown). The N-terminal sequence of the 83-kDa fragment was also from Ala-34, indicating that the second autolysis occurs at the C terminus of p94⌬ (data not shown).
To characterize further the autolytic activity of p94⌬ relative to that of full-length wild-type p94, p94:C129S, a good "intermolecular autolytic" substrate for p94, was proteolyzed by p94⌬ (24). Co-incubation of p94:C129S (1 g) and p94⌬ (0.1 g) resulted in the production of four major fragments with approximate molecular masses of 93, 91, 58, and 33 kDa, as shown in Fig. 2H. Interestingly, none of these fragments was recognized by anti-pNS antiserum, and the full-length 94-kDa protein detected by anti-pNS antiserum decreased rapidly in the reaction. The N-terminal residue of the 93-kDa fragment (Fig. 2H  (a)) was Ala-15 (Fig. 1A, arrow [1]), whereas the N-terminal residues of the 91-and 33-kDa fragments (Fig. 2H (b and d), respectively) were identical to that of N-terminally autolyzed p94⌬ (Ala-34, Fig. 1A, arrow [2]). These results indicate that p94⌬ very efficiently hydrolyzes the N terminus of p94:C129S, as well as that of p94⌬ itself. The N-terminal residue of the 58-kDa fragment (Fig. 2H (c)) was Glu-323 (Fig. 1A, arrow [3]), which corresponds to one of the previously determined autolytic sites in native p94 (31). The 33-kDa fragment was not detected by either the monoclonal antibody 12A2 or anti-pIS2 antiserum. These results indicate that p94:C129S is hydrolyzed by p94⌬ first at the N terminus and then in the region of IS1, which generates 58-and 33-kDa fragments corresponding to the C-and N-terminal parts of the 91-kDa fragment, respectively.
Substrates for p94⌬-Previously, we found that p94 causes a decrease in a 60-kDa protein in vivo, when expressed in COS cells (14). Peptide sequencing revealed that it corresponds to heat shock protein 60 (HSP60). Consistent with this, an in vitro assay showed that p94⌬ proteolyses HSP60 (Fig. 3A). More-  ; is is the intervening region between IgI80 and IgI81; and PEVK is the region rich in Pro, Glu, Val, and Lys. The vertical arrow indicates one of the m-calpain-susceptible proteolytic sites in connectin/titin. CN48 is the original N2A clone isolated by yeast two-hybrid screening using p94 as bait (17), and CN48-⌬1 to -⌬6 were generated from CN48. Each construct was co-transformed with bait plasmid bearing p94 into AH109, and binding was evaluated by growth on SD-LWA plates and ␤-galactosidase activity (shown in the right column). aar, amino acid region; N-term, N terminus; C-term, C terminus. C, proteolysis of N2A connectin/titin fragments by p94⌬, -calpain, or m-calpain. IgI80   Reactions in B (ϩp94⌬) were terminated by the over, HSP60 is proteolyzed by m-calpain much more rapidly than by p94⌬, suggesting that HSP60 is also a possible in vivo target of the conventional calpains. Although HSP60 is highly conserved from bacteria to eukaryotes (ϳ50% primary sequence identity between human and E. coli proteins), the E. coli HSP60 homologue, GroEL, is very poorly hydrolyzed by p94⌬. Its co-chaperonin, GroES, and combinations of GroEL and ES with or without ATP were also examined, but none was proteolyzed by p94⌬ (data not shown).
Several other proteins were incubated with p94⌬ to test whether they are substrates. Two different N2A fragments of connectin/titin, IgI80 -81 and IgI82-83, were hydrolyzed by p94⌬ as well asand m-calpains. The connectin constructs IgI80 -81 and IgI82-83 contain the two N-terminal and the two C-terminal Ig motifs of the N2A region, respectively. IgI82-83 includes a p94-binding site for connectin (Fig. 3B) (17). IgI80 -81 was more rapidly degraded than IgI82-83 by p94⌬ andand m-calpains, as shown in Fig. 3C. N-terminal sequencing of the degradation fragments of IgI80 -81 revealed that one of the proteolytic sites susceptible to calpain is on the N terminus of Gly-9434 (GenBank TM accession number NP_596869; Ref. 35; Fig. 3B, arrow) in I80. These results strongly suggest that not onlyand m-calpains but also p94 proteolytically degrade connectin/titin.
Protease-inactive m-calpain, m-calpain:C105S, was proteolyzed by p94⌬, as shown in Fig. 3D, whereas p94:C129S was not hydrolyzed by m-calpain (data not shown). It should be mentioned that anti-pIS2 detects p94⌬ because a part of the epitope is retained in p94⌬. Notably, the addition of the calpastatin domain 1 fragment (Fig. 3D, ϩCSTN) abolished mcalpainolysis by p94⌬ but not autolysis of p94⌬ or caseinolysis by p94⌬ (Fig. 2E), suggesting that the m-calpain-calpastatin complex acquires resistance to proteolysis by p94⌬ in the presence of Ca 2ϩ . A widely used fluorescent substrate for the conventional calpains, Suc-LLVY-MCA, was not significantly hydrolyzed by p94⌬ (data not shown).
These results show that some of the substrates for p94⌬ (and for p94) can be hydrolyzed by the conventional calpains as well (casein, HSP60, connectin/titin N2A fragments, and m-calpain: C105S). On the contrary, calpastatin and p94:C129S are hydrolyzed more efficiently and exclusively, respectively, by p94⌬ and p94.
Proteolysis of Calpastatin by p94⌬-Proteolysis of the calpastatin domain 1 fragment (14.1 kDa, residues 27-159 of human calpastatin, GenBank TM accession number NP_775085; Takara Shuzo catalog number 7316) by p94⌬ was examined by reversed-phase column chromatography. The calpastatin fragment was rapidly degraded proteolytically by p94⌬ into several fragments, as shown in Fig. 4, A and B, and the N-terminal sequences were determined. Six cleavage sites were identified (Fig. 4C, lines i-vi; four of these are shown in Fig. 4D), one of which is located close to the reactive site (line i). The others (lines ii-vi), if cleaved, result in the loss of one of the helical regions on the C-terminal side of the reactive site, which inter-acts with domain VI of the conventional calpains. All six cleavage sites are different from the cleavage sites of the caspases (Fig. 4D, open triangle) (36). These results indicate that the cleavage of calpastatin by p94 inactivates the calpain inhibitory activity of calpastatin. The most susceptible sites are on the N terminus of Ala-105 and Gly-124 (accession number NP_775085; ii and iv in Fig. 4, C and D), which are detectable in the fractions already appearing after 20 min of digestion (Fig. 4A, closed triangles). These sites both have Pro at the P3Ј position and a small amino acid (Ala and Gly) at the P1Ј position. Moreover, five of six cleavage sites in the calpastatin domain 1 fragment have two Pro residues between P2Ј and P8Ј. These features may reflect the proteolytic preference of p94. One of the autolytic sites in p94:C129S and p94⌬, the N terminus of Ala-15, also has Pro at the P3Ј position and a small amino acid (Ala) at the P1Ј position (Fig. 4C, N-1). However, there are no further significant similarities found between the amino acid sequences surrounding the cleavage sites in calpastatin and those for autolytic sites in p94:C129S and p94⌬ (Fig. 4C, N-2 and IS1-1-3) (31).
To test whether p94⌬ proteolytically degrades short peptides, a 27-mer peptide corresponding to the reactive site of calpastatin (Sigma C9181; Fig. 4D) was incubated with p94⌬. Although the efficiency of cleavage was very low (less than 5% after 120 min of incubation), the peptide was cleaved on the N terminus of Lys-99 (Fig. 4, C and D, vii). The site has Pro residues at the P1 and P2 positions but not in the PЈ positions, suggesting that short peptides are cleaved by p94⌬ differently from the cleavage of calpastatin domain 1 fragment, which is a much better substrate for p94⌬.
Assay System for p94 Utilizing Calpastatin as Substrate-As observed above, calpastatin is the best substrate for p94⌬ among the protein substrates examined in this study except for p94:C129S. Therefore, we developed an assay for p94 using the calpastatin fragment and fluorescent resonance energy transfer (FRET). When the purified substrate, Y-C-CSTN (Fig. 5A), was incubated with p94⌬, it was rapidly degraded into several fragments (Fig. 5B). Before the substrate was cleaved by p94⌬, the fluorescence spectrum of Y-C-CSTN excited at 434 nm, the CFP excitation wavelength, showed a peak at 527 nm, which is an emission wavelength of YFP, indicating that YFP is excited by the CFP emission by FRET (Fig. 5C, 0 min). The emission spectrum of Y-C-CSTN excited at 434 nm was measured during the proteolytic process. A time-dependent increase in CFP emission at 476 nm and a corresponding decrease in YFP emission at 527 nm was observed, demonstrating that the proteolysis of Y-C-CSTN by p94⌬ can be monitored as the change in the emission spectrum caused by the loss of FRET between those two fluorescent units (Fig. 5C). Various amounts of p94⌬ were tested for Y-C-CSTN digestion, and the ratio of YFP emission to CFP emission, 476:527 nm, was plotted for each time point (Fig. 5D). The increase in the fluorescence ratio 476:527 was time-and dose-dependent, and the activity of as little as 250 ng of p94⌬ could be assayed with an incubation of addition of 0.5 ml of 50 mM EDTA (pH 8.0), and the fluorescence emission spectra from 450 to 600 nm excited by 434 nm were scanned. D, dose-dependent fluorescent signals in this assay system. p94⌬ (0 -2 g) was incubated with 0. Reactions were carried out essentially as in E except that they were stopped by the addition of half a volume of 3ϫ SDS sample buffer. Western blot analysis of p94 (f.l.p94), p94⌬, and full-length p94:C129S (CS) with the antibody 12A2 and of -calpain (-calpain) with the antibody mAb3082 showed Ca 2ϩ -independent and Ca 2ϩ -dependent autolysis only for p94 and p94⌬, respectively. 45 min. -calpain produced no increase in the fluorescence ratio nor proteolysis of Y-C-CSTN over this period of time or in this range of protein concentrations (data not shown), suggesting that Y-C-CSTN is a p94⌬-specific substrate but not for the conventional calpains under our assay condition.
Activity of COS-expressed Full-length Wild-type p94 Detected by Y-C-CSTN-To test whether the assay described above can detect the proteolytic activity of full-length wild-type p94, as well as the activity of p94⌬, among 100 other co-occurring proteases in vivo, the proteolytic activity of p94 expressed in COS-7 cells was measured using Y-C-CSTN as substrate. COS cells are known to express considerable amounts of m-calpain and other cellular proteases, such as caspases. Ca 2ϩ -dependent proteolysis of Y-C-CSTN was detected in the lysates of cells expressing full-length wild-type p94 (Fig. 5E, f.l.p94) or p94⌬ (Fig. 5E, p94⌬) in the presence of protease inhibitors for major proteases including conventional calpains, caspases, and serine proteases. On the other hand, the lysates of cells expressing protease-inactive fulllength p94:C129S (Fig. 5E, CS) or -calpain (Fig. 5E, -calpain) showed no activity above background levels, i.e. above the value determined for cells transfected with empty vector, regardless of sufficient amount of the proteins detected (Fig. 5F, CS and -calpain). These results clearly indicate that our assay system specifically detects and distinguishes the protease activity of p94 in the presence of interfering proteases.
To our surprise, full-length wild-type p94 showed Ca 2ϩ -dependent calpastatinolytic activity (Fig. 5E, f.l.p94). The autolysis of recombinantly expressed p94, however, was Ca 2ϩ -independent (Fig. 5F, f.l.p94; gray arrowhead) as shown previously (14), which is consistent with the fact that most COS-expressed p94 had already disappeared at harvest. These results indicate that the autolysis of p94 does not require Ca 2ϩ , whereas calpastatinolysis occurs in a Ca 2ϩ -dependent manner, suggesting different substrate specificities for p94 in the absence and presence of Ca 2ϩ . DISCUSSION In this study, the enzymatic properties of one of the alternatively spliced variants of p94/calpain 3 were examined, to understand better the unique characteristics of p94 and their relevance to its physiological functions. Isolating significant amounts of the proteins is a prerequisite for in vitro enzymatic studies, and this has been almost impossible because of the very rapid autolytic activity of p94 (31). On the other hand, some natural splice variants of p94 expressed in skeletal muscle are somewhat stable. After several different isoforms were examined, p94⌬, which lacks both IS1 and IS2, was the most promising variant for large scale preparation. The structure of p94⌬ is almost identical to that of CL and mCL, except that it has the NS. It shows proteolytic characteristics similar to those ofand m-calpains, such as Ca 2ϩ dependence, insofar as IS1 and IS2 are involved in Ca 2ϩ -independent protease and autolytic activity, a hallmark of p94.
However, there are several traits distinguishing p94⌬ from the conventional calpains as follows: 1) p94⌬ shows specific caseinolytic activity much lower than those of the conventional calpains; 2) the optimal temperature for p94⌬ activity is higher (ϳ42°C) than those of the conventional calpains (25-30°C); and 3) p94⌬ has a lower pCa for half-maximal activity thanor m-calpains. At present, the possibility cannot be excluded that these properties originate from the lack of IS1 or IS2. It should be emphasized, however, that many features of p94⌬ are shared with full-length p94 as follows: neither is inhibited by but proteolytically degrades calpastatin; both exhibit proteolytic activity without 30K; both bind connectin/titin at the N2A and M line regions (16); both cut p94:C129S at identical sites; and both undergo autolysis in the presence of EDTA. Therefore, it is conceivable that the properties of p94⌬ de-scribed above reflect the nature of p94 and are related to the functions of p94 in the context of the physiology of skeletal muscle.
Stable and Active without 30K-The co-expression of 30K is required for recombinantand m-calpains and nCL-4 in the same Sf-9/baculovirus system (28,32,33,37,38). However, proteolytically active p94⌬ and inactive p94:C129S were isolated in the soluble and stable fraction without 30K. Furthermore, no interaction between p94⌬ and 30K was detected in the yeast two-hybrid system, as was also the case for p94 (17,16). These data strongly suggest that p94⌬ and p94 do not form heterodimers with 30K. It awaits further analyses on the tertiary structures of p94⌬ and p94 to determine whether they form homodimers which was predicted from the elution positions of p94:C129S and p94⌬ on the gel filtration column (see Ref. 31 and data not shown).
Autolytic Activity of p94 and p94⌬-Our previous results on the in vitro translation of p94 indicated that the autolysis of p94 proceeds in the presence of excess EDTA (14), which was also observed for the N-terminal autolysis of p94⌬. Branca et al. (39), however, reported Ca 2ϩ -dependent proteolytic fragmentation of recombinant p94. They isolated a substantial amount of full-length wild-type p94 using the Sf-9/baculovirus system. In our hands, however, p94:C129S was expressed abundantly and stably and was successfully purified almost to homogeneity using the same Sf-9/baculovirus system (Fig. 1B), whereas wild-type p94 expressed in the same system was detected only as a 55-kDa autolyzed fragment (data not shown). At present, no clear explanation is available for these discrepancies, but p94 may show Ca 2ϩ -dependent autolysis under certain conditions. Indeed, Ca 2ϩ -dependent hydrolytic activity was detected for full-length wild-type p94 as well as for p94⌬ using a calpastatin-based FRET substrate in this study.
Hydrolysis of p94:C129S by p94⌬, but not by m-calpain, strongly suggests that the substrate specificities of p94⌬ and p94 are the same. In addition to the autolytic site in IS1 (31), p94⌬ hydrolyzes p94:C129S at the N termini of Ala-15 and Ala-34. In this study, one of the intramolecular autolytic sites of p94⌬ was also demonstrated to be Ala-34. Ca 2ϩ -independent autolysis at the N terminus of Ala-34 generates a rather stable autolytic fragment of p94⌬ because no further autolysis occurred during the storage of p94⌬ for over a year (data not shown). Recently, Rey and Davies (34) reported that one of the autolytic sites in a recombinant protein corresponding to the protease domain of p94 was Ala-15, which is consistent with our result. They also reported Ala-45 and Thr-316 as autolytic sites, which were not identified in our experiments. It can be reasoned that the lack of domain III and regions thereafter in their construct exposed sites potentially susceptible to autolytic cleavage. Previously, we have observed that the Ca 2ϩ -independent autolytic activity of recombinant p94:exon6 Ϫ was much weaker than that of full-length p94, resulting in the stabilization of p94:exon6 Ϫ (16). 2 Therefore, it is inferred that the lack of exon 6, which constitutes most of IS1, contributes to the stability of p94⌬ by abrogating the pivotal autolytic site. Rey and Davies (34) also reported Ca 2ϩ -dependent autolysis of their protein, which can be ascribed to the lack of IS2 based on our previous observations (16). 2 Very recently, Fukiage et al. (40) reported the qualitative characterization of Lp82, another alternative splice variant of p94, which is specifically expressed in the lens. They showed that the lack of IS2 does not effectively stabilize the product when NS is replaced by AX1, a lens-specific N-terminal sequence (41). Considering these data together, we conclude that NS, IS1, IS2, and AX1, unique sequences specific to CAPN3 proteins, have independent functions and that different combinations of them confer specific characters, e.g. different autolytic activities, upon each splice variant.
p94-specific Assay toward a Diagnosis of LGMD2A-Because the loss of the substrate processing activity of p94 causes LGMD2A (24), we have focused on identifying its substrates. One of the major problems we faced was how to distinguish the activity of p94 from that of other proteases, including the conventional calpains, in a physiological context. The enzymatic properties determined for p94⌬ in this study showed that there are conditions favorable for p94 but not for the conventional calpains, such as high temperature and the presence of calpastatin. Consequently, we established a specific assay system for p94 and demonstrated its ability to identify COS-7expressed p94, as well as p94⌬.
In theory, 0.3 mg of muscle, which corresponds to about three 10-m cryostat slices of a muscle biopsy, contains at least 300 ng of p94 (42). Our system can assay as little as 250 ng of p94⌬. The activity of the COS-7-expressed p94 and p94⌬ detected in this study corresponds to a minimum of ϳ10 ng of protein. Therefore, we anticipate that the activity of endogenous p94 in tissues (biopsy samples) will be measurable using this assay and that this methodology will be applicable to the diagnosis of LGMD2A by screening for the loss of p94 activity.
Calpain Network and LGMD2A-The apparently Ca 2ϩ -independent proteolysis of p94:C129S and calpastatin is p94-specific, which is not observed for the conventional calpains (17). m-calpain:C105S was also degraded proteolytically by p94⌬ in a Ca 2ϩ -dependent manner, although p94:C129S was not degraded by m-calpain. These results imply, for the first time, that p94 has a certain role in regulating the conventional calpains directly and indirectly by proteolytic degradation of the calpains and calpastatin, respectively.
It has been reported that there are several biological contexts where several different protease systems keep cross-talking. For example, calpain and caspase proteolytic systems function synergically in apoptosis, where proteolytic inactivation of calpastatin by caspase facilitates calpain activation and calpain activates caspase by limited proteolysis (43). It has not yet been shown, however, whether there is a cross-talk between p94 and the conventional calpains, the so-called "calpain network" in skeletal muscle where the expression level of p94 predominates. It is already clear that the function of p94 cannot be completely compensated by other calpains because a defect of p94 function causes LGMD2A. Therefore, it is of our interest to investigate how p94 could intervene in the conventional calpain system, which would be compromised in LGMD2A.